With the introduction of piperacillin/tazobactam to the North American market, hospitals have been faced with the task of making a decision regarding its formulary role. In view of its broad spectrum of activity, piperacillin/tazobactam could be considered as a formulary alternative to imipenem. To evaluate the formulary feasibility of substituting piperacillin/tazobactam for imipenem, a comparative assessment of these agents in the empiric treatment of serious bacterial infections was undertaken at this tertiary care hospital. This trial was conducted as a randomized, double-blind, single-center study. Consenting adult patients (>16 years of age) who were prescribed imipenem were randomized to receive either 4 g of i.v. piperacillin/tazobactam or imipenem 500 mg of i.v. Q6H with or without concurrent antibiotics. Doses were adjusted according to renal function. There were no restrictions regarding the use of nonstudy antibiotics before and during the study period. Patients with beta-lactam allergies or meningitis or who had received greater than 72 h of previous imipenem therapy were excluded. Patients were evaluated at the end of treatment, at discharge, and at 30 days postdischarge. Endpoints included both clinical and microbiologic efficacy as well as drug toxicity. Over the 433-day study period, 360 imipenem treatment courses were initiated. Of these, 150 treatment courses (75 piperacillin/tazobactam courses and 75 imipenem courses) met study criteria and were subsequently randomized. The distribution of prescriber services for enrolled patients was similar to that for all patients receiving imipenem during the study period (p = 0.15). Also, there were no statistically significant differences in demographic parameters between enrolled and excluded patients. For those patients enrolled in the study, demographic characteristics, treatment course indication(s), and accompanying antibiotics were similar across treatment arms. The mean duration of study drug therapy was 7.7 days (SD, 6.2) for imipenem and 7.5 days (SD, 6.7)for piperacillin/tazobactam (p = 0.84). In the majority of cases, treatment discontinuation occurred as a result of a favorable treatment course outcome, stepdown to a narrower spectrum parenteral agent, or stepdown to an oral agent and did not differ between study drugs (p = 0.73). Clinical and microbiologic treatment course outcomes were also similar across treatment arms. Clinical outcome was deemed successful or improved for 68% of imipenem and 70% of the piperacillin/tazobactam treatment courses (p = 0.54). Fifty-three percent of treatment courses were microbiologically confirmed. Of the 58 courses that were assessed for microbiological outcome, 93% demonstrated successful eradication of the causative pathogens. There was no difference between study drugs (96% imipenem; 90% piperacillin/tazobactam; p = 0.61). The proportion of treatment courses with at least one adverse event was similar between the study drugs (p = 1.0). Nausea and/or vomiting were/was observed more commonly in the imipenem arm (p = 0.03). Discontinuation of therapy due to drug toxicity occurred in 16% of imipenem and 5% of piperacillin/tazobactam treatment courses (p = 0.06). There was no statistically significant difference between the mean treatment course cost for imipenem ($762; range, $55-$3192) versus piperacillin/tazobactam ($696; range, $79-$2967; p = 0.59). In summary, piperacillin/tazobactam seems to represent a suitable alternative to imipenem for several clinical indications including intraabdominal infections, pneumonia, febrile neutropenia, and skin/soft tissue infections in which the causative pathogens are susceptible. However, in view of the prevalence of multiresistant Gram-negative aerobic pathogens at this institution, we do not believe that imipenem can be removed from the drug formulary. In addition, at the currently studied dosing regimen, there seems to be no evidence of a direct cost advantage associated with